Comparison of Two Drainage Simulation Models Using Field Data

Abstract

ABSTRACT The transient, one-dimensional finite-difference model, SWATREN, was tested using field data from Aurora, North Carolina. The SWATREN model could not be used for conditions when the water table rose to the surface. Since the water table frequently rises to the surface on poorly drained soils in the humid regions of the United States, SWATREN was modified by the authors to enable the simulation of these high water tables. The simulation results were compared with the performance of the water management simulation model, DRAINMOD, which had previously been tested using the same data. The field site consisted of three drainage treatments with provision for subirrigation and controlled drainage. Rainfall intensity and water table elevations were continuously measured at the site from 1973 to 1977 (a total of approximately 4,600 site days). Input data, such as saturated hydraulic conductivity, soil water characteristic, and crop rooting depths used in the previous DRAINMOD study were used in the SWATREN model. Comparison of daily water table depths predicted with the modified version of the SWATREN model were in good agreement with measured values, with standard errors of estimate ranging from 8.9 cm to 29 cm. The DRAINMOD model had standard errors of estimate ranging from 7.5 cm to 19 cm. The average absolute deviation between predicted and measured water table depths for the 15 site years of data was 13.4 cm for SWATREN, as compared to 11.4 cm for DRAINMOD. Based on the results of this study, either the modified version of SWATREN or the DRAINMOD model can be used to simulate drainage response on soils with a fluctuating water table that may rise to the surface.